Reprinted From The Journal of The American Medical Association November 25, 1968, Vol. 206, pp. 1973-1977 Copyright 1968, by American Medical Association Lasker Award Genetic Memory Marshall Nivenberg, PhD enetic memory resides in specific molecules of deoxyribonucleic acid. The DNA alphabet consists of four letters, the bases, A, T, G, and C. The sequence of letters in a nucleic acid message corresponds to a sequence of the 20 amino acid species in protein. Two molecules of DNA interact with one another by hydrogen bonding between bases on opposite chains. As proposed by Watson and Crick,’ adenine pairs with thymine, and guanine with cytosine. Information is retrieved by tran- scribing the DNA message in the form of ribonu- cleic acid and then translating the RNA message into protein. Triplets are translated sequentially, from left to right. The information encoded in a nucleic acid tem- plate enables the reading mechanism to select one from many species of molecules, to define the posi- tion of the molecule relative to the previous mole- cule selected, and to define the approximate time of the event relative to previous events. Hence the nucleic acid functions both as a template for other molecules and as a biological clock. Although the genetic information is encoded in the form of a one-dimensional string, the polypep- tide products fold in a specific manner predeter- mined by the amino acid sequence. In effect, a complex, three-dimensional object is created by first fabricating a linear string of letters that folds upon itself, in a fairly specific manner. Probably the principle of unidimensional sculpturing could be used by man for the construction of certain kinds of objects. Often, one molecule of messenger RNA (mRNA) contains the information for many molecules of protein, so the RNA message must also contain in- formation for the initiation and termination of the polypeptide chain. The translation must be ini- tiated properly since selection of the first word also phases the translation of subsequent words. At least three enzymes are required for the initiation process, three additional enzymes for the forma- tion of the peptide bond and movement of the ribo- some along the message, and one or more enzymes for the termination of protein synthesis. In addi- From the National Heart Institute, Bethesda, Md. Presented as a 1968 Albert Lasker Basic Research Award Lecture at the New York University Medical Center. New York, Nov 26, 1968. Reprint requests to National Heart Institute, Bethesda, Md 20014. JAMA, Nov 25, 1968 © Vol 206, No 9 tion, specific enzymes are required for the synthesis and repair of DNA, for the synthesis of mRNA, aminoacyl-transfer RNA (AA-tRNA), and for the modification of tRNA and ribosomal RNA. The process of protein synthesis, illustrated diagram- matically and in highly abbreviated form, is shown in Fig 1. The chromosome of a relatively primative or- ganism, such as Escherichia coli, consists of approximately 3 million base pairs. Sufficient infor- mation is present to determine the sequence of 1 mil- lion amino acids in protein which is approximately the amount required for 3,000 species of protein. The human genome is 1,000 to 2,000 times larger than that of E coli, ie, information for the synthesis of 3 to 6 X 10° species of protein could be present. However, multiple copies of essentially the same gene frequently are stored; hence much information probably is redundant. The precise number of enzymes required for the storage, retrieval, and transmission of genetic in- formation has not been determined. Perhaps 200 species of protein would suffice. However, as much as 25% of the total protein synthesized by rapidly growing FE coli is utilized for the construction of new ribosomes. The Rate of Reading.—The average E coli ribo- some can read approximately 1,000 mRNA triplets per minute. The reading rate therefore is quite slow compared to a man-made computer, However, pro- tein is synthesized simultaneously at many sites within the cell. Escherichia coli with a generation time of 25 minutes contains approximately 15,000 ribosomes per chromosome. Hence, 15 million amino acids may be incorporated into protein per minute per chromosome. The RNA message usually is cov- ered by a train of ribosomes; hence one molecule of mRNA is translated simultaneously at different sites. A single molecule of mRNA may serve there- fore as a template for the synthesis of many mole- cules of protein. It seems likely though, that some species of mRNA are destroyed earlier than others. Some genes are transcribed more frequently than others. It is clear that retrieval of genetic informa- tion often is regulated selectively. Some regions of E coli DNA may be transcribed 1,000 times per generation; others may be transcribed only one or two times per generation. Deciphering the Genetic Language.—The experi- mental approaches that eventually led to the de- ciphering of the genetic code came from the study of in vitro synthesis of protein. The demonstration that mRNA is required for the in vitro synthesis of protein and that synthetic polynucleotides such as poly U serve as templates for the synthesis of polyphenylalanine provided a means of exploring many aspects of the code and the translation process.” Polynucleotides composed of different combinations of bases in random sequence were syn- thesized with the aid of polynucleotide phosphory- lase, discovered by Grunberg-Manago and Ochoa.° Synthetic mRNA preparations then were used to Genetic Memory—Nirenberg 1973 SATGCGAATGATCGAATGTCTGTTGTGCGCT...3 a a a a a a. JS TACGCTTACTAGCTTACAGACAACACGCGA,..5 2 tOR Fahy aaa asm (WETISULE ) ais an Panay nn StS 5 a} im eRs Dan io aie o a: AA ACTIVATION 1. The retrieval of genetic information is illustrated in a condensed and diagrammatic form. direct cell-free protein synthesis. The base content of the mRNA can be correlated with the amino acid content of newly synthesized protein. In this man- ner the base compositions of 53 RNA codons were assigned to amino acids.*” It was also found that three sequential bases in mRNA correspond to one amino acid in protein, that AA-tRNA is required for the translation of mRNA,* and that codons for the same amino acid sometimes require different species of AA-tRNA for their translation.” Analysis of the coat protein of mutant strains of tobacco mosaic virus provided evidence that triplets in mRNA are translated in a nonoverlapping fashion, because the replacement of one base by another in mRNA usually results in only one amino acid re- placement in protein.* Alternate codons for the same amino acid were shown in most cases to contain two bases in com- mon. Common bases were assumed to occupy the same base position of synonym triplets. Base Sequence of Codons.—Codon base se- quences were established in several ways: by di- recting in vitro protein synthesis with polyribonu- cleotides containing repeating doublets, triplets, or tetramers of known sequence as described by Kho- rana in the accompanying communication, and by stimulating the binding of aminoacyl-tRNA to rib- osomes with trinucleotides of known sequence.’ Since aminoacyl-tRNA binds to ribosomes prior to peptide bond formation, the process of codon recog- nition can be studied without peptide bond synthesis. A simple method for determining '*C-aminoacy]l- tRNA bound to ribosomes was devised that depends upon the selective retention of '*C-aminoacyl-tRNA bound to ribosomes by disks of cellulose nitrate; unbound '*C-aminoacyl-tRNA is removed by wash- ing. At the time that the trinucleotide template ap- proach to codon sequence was devised, most of the 64 trinucleotides had not been prepared. Elegant 1974 JAMA, Nov 25, 1968 ® Vol 206, No 9 Table 1.—The Genetic Code* UUU UCU UAU UGU PHE TYR cYsS uuc ucc UAC uGcc SER UUA UCA UAA TERM UGA TERM LEU UUG UCG UAG TERM UGG TRP cuu ccu CAU CGU HIS cuc ccc CAC cGCc LEU PRO ARG CUA CCA CAA CGA GLN CUG CCG CAG CGG AUU ACU AAU AGU ASN SER AUC ILE ACC AAC AGC THR AUA ACA AAA AGA LYS ARG MET, AUG MET ACG AAG AGG GUU GCU GAU GGU ASP GUC Gcc GAC GGc VAL ALA GLY GUA GCA GAA GGA GLU MET, GUG GCG GAG GGG *Nucleotide sequences of RNA codons were determined by stimu- lating binding of E coli AA-tRNA to E coli ribosomes with trinucleotide templates. F-Met corresponds to N-formy!-Met-tRNA, the initiator of protein synthesis. TERM corresponds to terminator codons. chemical methods for oligoribonucleotide synthesis, devised by Khorana and his colleagues, are de- scribed in the accompanying communication. We have employed enzymatic methods for oligoribonu- cleotide synthesis. Leder and co-workers showed that primer-dependent polynucleotide phosphory- lase, in the presence of a dinucleoside monophos- phate primer and nucleoside diphosphate, catalyzes the synthesis of oligonucleotides of low chain jength’®; a similar method was also reported by Thach and Doty.’* Another enzymatic method for oligonucleotide synthesis, reported by Bernfield,'*''’ is based upon the demonstration by Heppel et al’* that RNase A catalyzes the synthesis of oligonu- cleotides from pyrimidine-2’,3’-cyclic phosphate moieties in the presence of mononucleotide or olig- onucleotide acceptors. The 64 trinucleotides were synthesized and as- sayed for template specificity in stimulating bind- ing of EF coli aminoacyl-tRNA to ribosomes.’°’* A summary of the code is shown in Table 1. Almost all triplets were found to correspond to amino acids. In most cases, synonym codons differ only in the base occupying the third position of the tyriplet. Thus synonym codons are systematically related to one another. Only four unique patterns of degenera- cy were found, each pattern determined by the bases that occupy the third positions of synonym triplets. Patterns of alternate third bases are as fol- lows: () G @) U=C (3) A=G (4) U=CrHA Genetic Memory—Nirenberg A fifth pattern, U = C = A =G, was found also, but may be formed by combining two or more simpler patterns such as [(U =C) + (A =G)] or [((U =C=A) + (G)}. Codons specifying the initiation of protein syn- thesis differ in that alternate bases occupy the first rather than the third position of the codons. For example, N-formyl-Met-tRNA responds to AUG and GUG. Three triplets, UAA, UAG, and UGA, serve as terminator codons. Hence the degeneracy pattern again is unusual (discussed under Punctua- tion). One consequence of logical degeneracy is that mutations resulting from the replacement of one base pair in DNA by another often do not result in the replacement of one amino acid by another in protein. Hence, many mutations are “silent.’? The code appears to be arranged so that the effects of error often are minimized. Amino acid replace- ments in protein that result from an alteration of one base per triplet can be derived from Table 1 by moving horizontally or vertically from the amino acid in question, but not diagonally. Punctuation Codon Positinn.—Each triplet can occur in three structural forms: as a 5’-terminal-, 3'-terminal-, or internal-codon. Substituents attached to terminal or internal ribose hydroxyl groups can influence the template properties of codons profoundly. Relative template activities of oligo U preparations, at limit- ing oligonucleotide concentrations, are as follows: p-5'-UpUpU > UpUpU > CH;0-p-5'UpUp > UpUpU-3’-p > UpUpU-3’-p-OCH; > UpUpU-2’, -3’-cyclic phosphate. Trimers with (2’-5’) phospho- diester linkages, (2’-5')-UpUpU and = (2’-5’)- ApApA, do not serve as templates for phenylala- nine- or lysine-tRNA, respectively. The relative template efficiencies of oligo A preparations are as follows: p-5'-ApApA > ApApA > ApApA-3’-p > ApApA-2’-p.”” Many enzymes have been described that catalyze the transfer of molecules to or from terminal hy- droxyl groups of nucleic acids. It is possible there- fore that modifications of terminal hydroxyl groups sometimes regulate the reading of RNA or DNA. Initiation.—-Two species of methionine-tRNA are found in E coli; one species, Met-tRNA, is convert- ed enzymatically to N-formyl-Met-tRNA,,’® and functions as an initiator of protein synthesis in ex- tracts of EF coli in response to the codons AUG or GUG; the other species, Met-tRNA,,, does not ac- cept formyl groups and responds only to AUG.'*:° Translation of mRNA is initiated near the 5’- terminus of the RNA and proceeds three bases at a time toward the 3’-terminus. The first amino acid to be incorporated into protein is the N-terminal amino acid; the C-terminal amino acid is the last to be incorporated. At least three nondialyzable factors are required for the initiation of protein synthesis.”**° However the reactions have not been clarified fully. It seems JAMA, Nov 25, 1968 ® Vol 206, No 9 probable that one factor, the C protein, is required for the attachment of the 30S ribosomal subunit to the 5’-terminus of the nascent chain of mRNA prior to the detachment of the mRNA from the DNA template.”? Another factor is required for bind- ing of N-formyl-methionyl-tRNA to the 30S ribo- somal subunit in response to AUG or GUG. Termination.—Results obtained by Stretton and co-workers” and by Garen®* demonstrate that UAA, UAG, and UGA are terminator-codons. Capecchi has reported that a protein, termed the release fac- tor, is required for terminator-codon dependent release of polypeptides from ribosomes.”° Terminal events in protein synthesis have recent- ly been studied with trinucleotide codons.” Initiator and terminator trinucleotides sequentially stimu- late N-formyl-methiony]-tRNA binding to ribo- somes and the release of free N-formyl-methionine from the ribosomal-intermediate. The release factor and a terminator trinucleotide is required for this reaction. The release factor has been fractionated into two components; R1, which corresponds to the terminator codons UAA and UAG; and R2, which corresponds to UAA and UGA.” The specificity of R therefore is related to the codon. These results suggest that terminator codons may be recognized by release factors. However, the mechanism of ter- mination remains to be clarified, and it is certainly possible that terminator-codons are recognized by components that have not been detected thus far. Mechanism of Codon Recognition Cells often contain multiple species of tRNA for the same amino acid. Soon after the code was found to be degenerate, the specificity of separate species of tRNA'™ for codons was examined. Randomly ordered poly UG and poly UC preparations are templates for different species of Leu-tRNA.” Thus alternate codons for the same amino acid some- times are recognized by different species of tRNA. When base sequences of synonym codons were es- tablished, it became abundantly clear that synonym codons are logically related to one another. Since only a few general degeneracy patterns were found, each pattern was thought to represent a general mechanism for codon recognition. Evidence that one molecule of AA-tRNA can re- spond to two kinds of codons was obtained by show- ing that >99% of the available molecules of ‘*C- Phe-tRNA bind to ribosomes in response to poly U, and >65% of the molecules also bind in re- sponse to UUC.*® Hence >65% of the Phe-tRNA molecules respond both to UUU and UUC. Addi- tional evidence was obtained by fractionating AA- tRNA and determining the responses of the sep- arated fractions of **C-AA-tRNA to trinucleotide codons. Results obtained thus far in our laboratory with purified fractions of AA-tRNA from E coli, yeast, and guinea pig liver are summarized in Fig 2. It is clear that one species of tRNA may recog- Genetic Memory—Nirenberg 1975 Table 2.—Alternate Base Pairing™ tRNA mRNA Anticodon Codon U A G c G A U G c U I U c A Alternate base pairing. The base in a tRNA anticodon shown in the left-hand column forms antiparallel hydrogen bonds with the base(s) shown in the right hand column, which usually occupy the third position of alternate mRNA codons. Relationships are “wobble’’ hydrogen bonds suggested by Crick,30 nize 1, 2 or 3 synonym cod- ons that differ only in the base occupying the third position of the codon. Five unique patterns of degeneracy were found, each pattern determined by alternate third bases of synonym triplets recognized by a tRNA species. Patterns of alternate third bases of synonym codon sets are as fol- lows: (1) (2) (3) (4) (5) A Gran Cc G Cc G Crick proposed a mechanism that would enable a base in the tRNA anticodon to pair with alternate bases occupying the third position of synonym mRNA codons.*° By changing positions slightly, that is, by wobbling, bases in the appropriate posi- tion of the tRNA anticodon form alternate pairs with bases occupying the third position of synonym mRNA codons. Antiparallel Watson-Crick hydro- gen bonds form between the first and second bases of the mRNA codon and corresponding bases in the tRNA anticodon and wobble hydrogen bonds form between bases occupying the third positions of synonym mRNA codons and a corresponding base in the tRNA anticodon as shown in Table 2. Hence, U in a tRNA anticodon pairs with A or G in the third position of synonym mRNA codons; C pairs with G; G pairs with C or U; and I pairs with U, C, or A. Additional evidence supporting this mech- anism of codon recognition stems from the elucida- tion of base sequences of tRNA anticodons. The data are fully consistent with wobble base pairing. In summary, degeneracy patterns for amino acids observed with unfractionated AA-tRNA often re- sult from recognition of several codons by a single Species of tRNA and from the presence of multiple species of tRNA for the same amino acid that respond to different sets of codons. Universality Although the results of many studies indicate that the genetic code is largely universal, the fidelity A U 1976 JAMA, Nov 25, 1968 @ Vol 206, No 9 corresponds to Met Release factors 1 and UGA, respecti 2. Responses of purified AA-tRNA fractions to trinucleotide codons. Joined sym- bols adjacent to codons represent synonym codons recognized by a purified AA-tRNA fraction from @ E coli, A yeast, and JJ guinea pig liver. The number between sym- bols represents the number of redundant peaks of AA-tRNA found responding to that set of codons. The open symbols represent ambiguous AA-tRNA responses; MET; corresponds to N-formyl-Met-tRNA, the initiator of protein synthesis; MET,, “tRNA; TERM corresponds to termination of protein synthesis. and 2, rather than RNA, correspond to UAA and UAG, or UAA vely (a signifies uncertain). of translation can be altered in vivo and in vitro by altering components or conditions required for protein synthesis. The extent of such alterations was examined by studying the fine structure of the code with tRNA from different organisms. Almost identi- cal translations of nucleotide sequences of amino acids were found with bacterial, amphibian, and mammalian aminoacyl-tRNA. However, E coli tRNA did not respond detectably to certain codons. Therefore aminoacyl-tRNA preparations were frac- tionated by column chromatography and responses of tRNA fractions to trinucleotide codons were determined.** A summary of the results is shown in Fig 2. Many “universal” species of aminoacyl-tRNA were found, however seven species of mammalian tRNA were not detected with E coli preparations; conversely, five species of tRNA from E coli were not found with mammalian preparations. The re- sults also suggest that some organisms contain little or no aminoacyl-tRNA for certain codons (AUA, AGA, or AGG). The remarkable similarity in codon base se- quences recognized by bacterial, amphibian, and mammalian AA-tRNA suggests that most, perhaps all, forms of life on this planet use essentially the same genetic language. The code probably evolved more than 5 X 10° years ago. It is possible that some species-dependent dif- ferences in the codon recognition apparatus serve as regulators of protein synthesis. The possibility that embryonic differentiation may be dependent upon changes in codon recognition remains to be ex- plored. At the present time, the biological conse- Genetic Memory—Nirenberg quences of a modifiable translation apparatus are largely unknown. Fidelity.—Since multiple species of tRNA for the same amino acid often recognize separate sets of codons, the synthesis of two proteins with similar amino acid compositions may require different spe- cies of tRNA. Some codons probably occur more frequently in mRNA than others for the same amino acid. Most codons probably are translated with little error (0.1% to 0.01%). However, with some codons the level of error may be as high as 50%. Therefore, the accuracy of codon translation can vary at least 5,000-fold. Errors usually are specific ones, be- cause two out of three bases per codon often are translated correctly. The code seems to be ar- ranged so that the consequences of error often are minimized. The biological significance of a flexible, easily modified codon translation apparatus is not known. One intriguing possibility is that the codon recog- nition apparatus is modified in an orderly, predicta- ble way at certain times during cell growth and differentiation and that such modifications selec- tively regulate the kinds and amounts of proteins synthesized. In accord with this hypothesis, many factors have been found that influence the rate and the accuracy of protein synthesis in vitro. In addi- tion, one may also consider the structural hetero- geneity of components required for protein synthe- sis. For example, it seems probable that tRNA may be extensively modified by enzymes after the tRNA polynucleotide chain has been synthesized. Since tRNA contains many trace bases, a spectrum of intermediates probably exists for each species of tRNA. Whether such reactions play a role in regu- lating gene expression remains to be determined. One intriguing possibility is that infection of a cell by a virus may result in the production of a factor that modifies tRNA and, in consequence, alters the rate of synthesis of mRNA or protein. It is clear that the translation apparatus of the cell will accept and follow in robot-like fashion any instructions written in the appropriate molecular language. Since the language has been deciphered, the informational properties of the genetic message can be defined in terms of molecular structure. It seems probable that synthetic messages will even- tually be used to program cells and their descen- dants. References 1. Watson, J.D., and Crick, F.H.C.: Molecular Structure of Nucleic Acids: A Structure for Deoxyribose Nucleic Acid. Nature 171:737-738 (April 25) 1953. 2. Nirenberg, M.W., and Matthaei, J.H.: The Dependence of Cell-Free Protein Synthesis in E coli Upon Naturally Occurring or Synthetic Polyribonucleotides, Proc Nat Acad Sci USA 47: 1588-1602 (Oct 15) 1961. 3. Grunberg-Manago, M.: Ortiz, P.J.; and Ochoa, S.: Enzymic Synthesis of Polynucleoctides, Biochim Biophys Acta 20:269-285, 1956. 4. Speyer, J.F.. et al: Synthetic Polynucleotides and Amino Acid Code, Cold Spring Harbor Symp Quant Biol 28:559-567, 1963. 5. Nirenberg, M.W., et al: On Coding of Genetic Informatien, Cold Spring Harbor Symp Quant Biol 28:549-557, 1963. 6. Nirenberg, M.W.: Matthaei, J.H.; and Jones, O.W.: An In- termediate in the Biosynthesis of Polyphenylalanine Directed by Synthetic Template RNA, Proc Nat Acad Sci USA 48:104- 109 (Jan 15) 1962. 7. Weisblum, B.; Benzer, S.; and Holley, R.W.: A Physical Basis for Degeneracy in the Amino Acid Code, Proc Nat Acad Set USA 48:1449-1454 (Aug) 1962. 8. Wittmann, H.G., and Wittmann-Liebold, B.: Tobacco Mosaic Virus Mutants and the Genetic Coding Preblem, Cold Spring Harbor Symp Quant Biol 28:589-595, 1963. 9. Nirenberg, M.W., and Leder, P.: RNA Codewords and Pro- tein Synthesis: The Effect of Trinucleotides Upon the Binding of sRNA to Ribosomes, Science 145:1399-1407 (Sept 25) 1964. 10. Leder, P.; Singer, M.F.; and Brimacombe, R.L.C.: Syn- thesis of Trinucleoside Diphosphates With Polynucleotide Phos- phorylase, Biochemistry 4:1561-1567 (Aug) 1965. ll. Thach, R.E., and Doty, P.: Enzymatic Synthesis of Tri- and Tetranucleotides of Defined Sequence, Science 148:632-634 (April 30) 1965. 12. Bernfield, M.R.: Ribonuclease and Oligoribonucleotide Synthesis: I. Synthetic Activity of Bovine Pancreatic Ribo- nuclease Derivatives, J Biol Chem 240:4753-4762 (Dec) 1965. 13. Bernfield, M.: Ribonuclease and Oligoribonucleotide Syn- thesis: II. Synthesis of Oligonucleotides of Specific Sequence, J Biol Chem 241:2014-2023 (May 10) 1966. 14. Heppel, L.A.; Whitfield, P.R.; and Markham, R.: Biochem J 60:8, 1955. 15. Nirenberg, M. et al: RNA Codewords and Protein Syn- thesis: VII. On the General Nature of the RNA Code, Proce Nat Acad Sct USA 53:1161-1168 (May) 1965. 16. Sél, D., et al: Studies on Polynucleotides: XLIX. Stimula- tion of the Binding of Aminoacyl-sRNA’s to Ribosomes by Ribotrinucleotides and a Survey of Codon Assignments for 20 JAMA, Nov 25, 1968 © Vol 206, No 9 Amino Acids, Proc Nat Acad Sci USA 54:1378-1385 (Nov) 1965. 17. Rottman, F., and Nirenberg, M.: RNA Codons and Pro- tein Synthesis: XI. Template Activity of Modified RNA Codons J Molec Biol 21:555-570 (Nov) 1966. 18. Marcker, K., and Sanger, F-.: N-Formyl-methiony]-S- RNA, J Molec Biol 8:835-840 (June) 1964. 19. Clark, B.F.C., and Marcker, K.A.: The Role of N-Formyl- methionyl-sRNA in Protein Biosynthesis, J Molec Biol 17:394- 406 (June) 1966. 20. Kellogg, D.A., et al: RNA Codons and Protein Synthesis: IX. Synonym Codon Recognition by Multiple Species of Valine-, Alanine-, and Methionine-SRNA, Proc Nat Acad Sci USA 55: 912-919 (April) 1966. 21. Anderson, J.S., et al: GTP-Stimulated Binding of Initia- tor-tRNA to Ribosomes Directed by /2 Bacteriophage RNA, Nature 216:1072-1076 (Dec 16) 1967. 22. Nomura, M., and Lowry, C.V.: Phage F2 RNA-Directed Binding of Formylmethionyl-TRNA to Ribosomes and the Role of 30S Ribosomal Subunits in Initiation of Protein Syn- thesis, Proc Nat Acad Sci USA 58:946-953 (Sept) 1967. 23. Iwasaki, K., et al: Translation of the Genetic Message: VII. Role of Initiation Factors in Formation of the Chain Ini- tiation Complex With Escherichia coli Ribosomes, Arch Bio- chem 125:542-547 (May) 1968. 24. Stretton, A.O.W.; Kaplan, S.; and Brenner, S.: Nonsense Codons, Cold Spring Harbor Symp Quant Biol 31:173-179, 1966. 25. Garen, A.: Sense and Nonsense in Genetic Code, Science 160:149-159 (April 12) 1968. 26. Capecchi, M.R.: Polypeptide Chain Determination in Vi- tro: Isolation of a Release Factor, Proce Nat Acad Sci USA 58: 1144-1159 (Sept) 1967. 27. Caskey, C.T., et al: Sequential Translation of Trinucleo- tide Codons for the Initiation and Termination of Protein Syn- thesis, Science 162:135-138 (Oct 4) 1968. 28. Scolnick, E., et al: Release Factors Differing in Specificity for Terminator Codons, Proce Nat Acad Sci USA, to be published. 29. Bernfield, M.R., and Nirenberg, M.W.: RNA Codewords and Protein Synthesis: The Nucleotide Sequences of Multiple Codewords for Phenylalanine, Serine, Leucine, and Proline, Science 147:479-484 (Jan 29) 1965. 80. Crick, F.H.C.: Codon—Anticodon Pairing: The Wobble Hypothesis, J Molec Biol 19:548-555 (Aug) 1966. 31. Caskey, C.; Beaudet, A.; and Nirenberg, M.: RNA Codons and Protein Synthesis: 15. Dissimilar Responses of Mammalian and Bacterial Transfer RNA Fractions to mRNA Codons, J Molec Biol, to be published. Genetic Memory—Nirenberg 1977